Posted by barb on Oct 6, 2004 in Thesis/Grad Life
The funding situation seems to have been ironed out for my thesis. I’ll be working 75% time as a contractor and 50% time as a grad student. Yup. 125% time…50 hours per week. Well, I suppose I already put that in with the evening and weekend babysitting I do for my data runs. The bad news is that my contractor vacation time accrues at only 50% time (true for anyone working anything less than 40 hours per week). The good news is that all of my other benefits are still in place.
Not sure if I’m happy with this, though. I certainly appreciate all the effort various people have made to make this happen, the upshot of which is that I don’t loose much of my salary. However, now I’m split between two companies, have to document that I work more, lose vacation time, and my tax withholdings will be all messed up (a couple years ago I came very close to owning money due to working for two different companies).
The worst part is that I’m not even sure that a PhD is what I want. I’m at that point where if I quit, I’ve wasted 2 years of my life, but if I continue, I’ll be miserable for the next 2-3 years. I’m not sure which is worse. Oh, and there’s the fact that I won’t actually use this degree when I’m finished that makes it all the more enticing.
I’m going to quietly put my resume up on monster.com and see if there are any good nibbles…
Posted by barb on Sep 29, 2004 in Thesis/Grad Life
I suppose I should have been expecting something bad to happen — it seems that any time something really good happens in my life, I get a swift kick in the gut a week or two later to remind me that life can suck, too.
Kim had warned me a few weeks ago that she might have trouble finding funding for my thesis work, but that the lab should be able to cover it if she couldn’t find money from her own grants. Well, her grants ran out, and the lab will have to start covering my grad salary, but they won’t pay me at the contractor salary that I’ve been making. The upshot? I’ll have to take a pay cut of 25% if I want to continue my thesis work.
So that’s now the big question. Do I want to continue my thesis work?
My intial gut reaction was, no, of course not. I quit my grad work in 2001 because I no longer wanted to life on a grad student’s salary — it was too much to have to constantly worry about how I was going to pay all my bills each month (answer: they didn’t always get paid…)
But, on the other hand, I’ve come so far, and I am at least in a different situation than I was before when I first quit. First of all, I’ll still have “real pay” for half of my work, so I won’t be dropping into the absolute pit of grad-student life again. Secondly, I have a husband who wants me to finish (if I want to finish), and will help out with expenses and cut our budget where we need to in order to make it work.
Back to the other side, though, I’ve been finding my thesis work unsatisfying and a slog for the last year. I find that I don’t necessarily care about the results, and I’ve been wondering if my choice to return was really wise. I’ve discovered that while my reason for leaving may have been primarily financial, it seems that it was also based on a true dissatisfaction with my choice of career.
I’ll regret it if I leave now.
I’ll be miserable if I stay.
Posted by barb on Sep 3, 2004 in Thesis/Grad Life
I had a great meeing with Kim and Chris today about my thesis. First, I convinced them that the oddness that we’d been seeing in MCG -6-30-15 is, in fact, instrumental. And, this convinces us further of the importance of using spectral analysis, especially for long-term studies of XTE sources.
Then I showed them my results from flux-selected fits of the MCG -6-30-15 data from the same data set as a published set. My results did not exactly match theirs, but Chris argued that he had never thought that theirs made a lot of sense, from a physical standpoint. The reasons my results are better stem from the improvements of the background and response modeling that the XTE team has done in the intervening years. I can have more data points than they did, but still retain similar error bars. In addition, I can use a bit more of the band-pass than they used, which helps pin down some of the model parameters.
Finally, I showed them the results of time-selected spectra from MCG -6-30-15 and NGC 5548 from my current pipeline. I had fixed a couple major problems in the pipeline, so these results look much, much better. In fact, they look good enough that Chris suggested that I just start plowing through the mountains of data that I have now. Yikes! This is what I’ve been working toward for the last year, so it’s quite exciting…
While the data is running, I need to start thinking about an outline for my thesis, as I see it now. I also need to start reasearching the sources I’m analyzing, and I still need to do a few things to help understand that oddness in MCG -6-30-15 that we’d seen so that I can write it up for a chapter in my thesis.
Posted by barb on Aug 26, 2004 in Thesis/Grad Life
I’m such an idiot sometimes that it’s a wonder that I can acutally dress myself in the morning.
Two idiotic things have happened in the last week with my thesis. One, I discovered that the response files that I’ve been making for the spectra are wrong for more than half the spectra, and two, I’ve been losing some of the latest data because I haven’t updated my SAA (South Atlantic Anomaly) history file.
The response matrix problem arose because I have a script that automatically makes the response files. The FTOOL needed to make the response requires the numbers of the PCUs used to extract the spectrum. I adjust these over time, since PCU #1 is only on intermittantly after a certain date, and PCU #0 lost it’s xenon later at a different date. It’s best to use as many PCUs as available for the data, so I use #1 and #0 before their respective malfunctions, but not after. In my response-generating script I had been defining which PCUs were used, but not actually feeding this to the FTOOL making the response. Idiot!
The SAA problem came about because I have a file in the “aux” directory for each observation that the reduction script (the rex FTOOL) looks at. The default requirement for data to be considered “good” is that the data be at least 30 minutes from the last SAA passage. The file that I’ve been using ended long before the latest data sets that I’ve downloaded, so all of the data has been rejected from these data sets. Idiot!
Anyway, I’ve since changed the response-generating script to feed the pcarsp FTOOL the acutal PCUs used, and I’ve updated the SAA history file in the aux directories.
Let’s try to avoid any further idiocy, boys and girls. Shall we?
Posted by barb on Aug 10, 2004 in Thesis/Grad Life
Frequently, the pressures of grad school would manifest themselves as totally insane conversations between me and other grads. Perhpas the pinnacle of this behavior was a discussion I had with Jocko about the “window cow”.
My university has one of those cows with a hole in it’s stomach that can be accessed through a “window” on it’s side. The cow is usually grazing in a pasture that is right on the path between the parking lot and the astronomy building. (Yup, there’s even a barn right across from the astronomy building, but while it’s cute to see the tiny goats in the spring, it does not produce a “country fresh” odor that you’d want in your house.) This “window” is mounted on the side of the cow with a support that protrudes out from the cow’s side by an inch or so.
One day, as Jocko and I were walking out to the parking lot, a thought struck me, and, of course, I articulated this thought. “So, how would they breed the window cow? I mean, couldn’t a bull just rip that window right out?”
Posted by barb on Jul 30, 2004 in Thesis/Grad Life
Met with Kim and Chris at long last this afternoon. The last time we all met was in April, and while I’ve gone to Kim a couple times with questions in the interim, it was good to meet with both of them and talk about my thesis again.
We talked a bit about my troubles with the deadtime correction with the Crab data. We’re not convinced that the Crab is the right source to be using, and Chris is going to e-mail another scientist to see if he has suggestions for better galatic sources to use. Unfortunately, those sources are still likely to require deadtime corrections. We also grabbed another scientist in the lab, Craig, who has done a lot of work with XTE data, and he was able to answer a few questions about the deadtime correction.
One interesting thing that came out of our talk with Craig was that XTE Epoch 5 does not correspond with a gain change, but with PCU 0 blowing it’s propane layer. This means that during Epoch 5, even though PCU 0 is still generating data, it may not have the spectral response that we are expecting, and may not be well-modelled. This means that I will have to go back and re-extract the Epoch 5 data for both the Crab and MCG -6-30-15 with PCU 0 excluded.
We also talked a bit about my work with the pipeline results. Unfortunately, it looks like the blind-fitting results are not coming back as reasonable. This is a problem because if I start to “massage” the spectral fits, our own biases can be easily introduced. We’d really like to say that blind fitting returns reasonable results, but they don’t, and this is largely because the bandpass of XTE and the degeneracies of the reflection models are working against each other.
At least I have a few directions to go in right now, and I’m looking forward, once again, to my thesis work over the next few weeks.
Posted by barb on Jul 14, 2004 in Thesis/Grad Life
I set aside my work with the Crab data, since I was just getting frustrated instead of making any progress. I’ll need to go back and figure out how to deal with the deadtime correction in the light curves, but that will wait until next week.
In the mean time, my work with MCG-6-30-15 has shown that there is a definite difference in results when extracting flux-resolved spectra versus time-resolved spectra. So far, my data pipeline has only done time-resolved spectra, but we’re not sure anymore that this is the right approach.
My project the last week has been to add flux-selected spectra extraction to the pipeline. I thought it would be difficult, and I do have some more tweaking to do that might turn out to be harder than I’d like, but my initial implementation did not take as long as I thought it might.
Currently, the pipeline treats each “observation” (as defined by the RXTE data) separately. It cuts the lightcurve into 0.25 cts/second slices, and then bins those up until there are at least 100,000 net photons in a bin. This ensures that there is enough signal to get a decent fit of the spectra to various models, but hopefully minimizes spectra that cut across flux states.
I’d like to improve this by making all of the flux bins equal across the entire data set (i.e. so that the flux bins from observation to observation are taken from the same count rate ranges). I can’t currently think of any way to do this easily, but I’ll ponder it for the next week or so, and maybe something will come to mind.
Posted by barb on Jul 13, 2004 in Thesis/Grad Life
One fun tradition that I started for myself during grad school was to do Christmas baking early in December. Anyone who has been in college will probably recall that early December is generally a particularly busy time. It marks the last two weeks, or so, of regular classes, when the professors realize that they have not nearly covered the amount of material they’d been planning to cover. This means that in addition to the normal end-of-the-semester papers , exams and prep for finals, the amount of material and homework presented during each class increases by 50-100%.
This was the time of the semester that I wanted to do my Christmas baking.
Posted by barb on Jun 29, 2004 in Thesis/Grad Life
The Crab data is winning, too.
A few weeks ago, before the proposal was in our scopes, Kim had suggested that I use the Crab data as a way to check our results with MCG -6-30-15. This sounded like a reasonable idea that shouldn’t take too much work, since I’m familiar with reducing RXTE data (more than familiar, really, at least for extra-galactic sources). However, I had not factored in the complications of dealing with a galactic source. Why is this a problem? Since the source is so much closer, the telescope gets a much larger signal from it, and then complications of instrument dead-time arise*.
My first results looked weird, and Kim later pointed out that this was likely because of a dead-time correction. So, I looked up how to correct for dead-time on the RXTE data anlaysis pages, and found an article on correcting spectra. However, there was nothing on how to correct light curves. I decided to see how the results would look using count rates derived from integrating under the spectrum, which I finally finished yesterday. Unfortunately it really doesn’t tell me much. My previous plots had data binned to 512 seconds, which produced a few thousand data points. With these spectrum-derived points, there are only about 20 points. Not a good comparison, really.
Now I’m pondering how to correct the lightcurves directly, so I can get my volume of data points back up. In the mean time, I’m going to start writing scripts to produce flux-selected spectra in the data pipeline. I need to consider how best to do this to balance the need for lots of photons per spectrum while trying not to have the spectrum cross into more than one flux state.
* Detectors take a finite time to recover from detecting an incomming photon. If more photons come in during this recovery period, they are not fully registered, so the count rate measured is actually a bit less than the actual count rate. For extra-galactic X-ray sources this isn’t really a problem because the photons don’t come in fast enough, in general, to arrive during the dead-time from a previous photon. However, this can be a real problem for galactic sources.
Posted by barb on Jun 23, 2004 in Thesis/Grad Life
I’ve been working as a support scientist for three and a half years here at Goddard, and in those three years I have assisted with four different proposals, including the one that’s due Friday. In all of the other cases, the proposal has been furiously edited and re-edited until the last possible moment. I’ve found myself making the required 15 or 20 copies at 10 AM on the due date, frantically trying to find a working copier or searching offices for a working stapler. This only works because we are located in Greenbelt Maryland, a mere 15 miles (or so) from NASA Headquarters in DC where the proposals must be delivered before 4:30 PM on the due date. Someone from the lab would take the proposals for several people down to HQ, squeaking in the door at 4:20 PM.
Let me just say that I’m the kind of person who likes to have things done early. This last-minute-scramble drives me crazy. I would rather have the proposal in the bag with a day or two to spare, and send it FedEx overnight so that I don’t have to worry about navigating downtown DC.
This time around, I finally got my way! The only reason this happened is because Kim is out of town (left on Monday, so her contribution had to be finished by then) and Chris is busy working on his own proposals (though he, too, seems to be out of town, so I’m not sure how that’s working). So, I was able to finish the proposal yesterday with a couple careful read-throughs (and proofreading by Andrew), and a half-hour in front of the copier. The proposal is now away! And it’s not due until Friday!